Change tensorimpl opencv `future_std::span<cv::Mat>` to `cv::Mat`

Why ? TensorImpl_opencv attribute future_std::span<cv::Mat> mData; is not relevant.

MR Content :

• Change to cv::Mat mData;
• Adapt functions of TensorImpl_opencv
• Adapt tests
• Change the opencv::imread flag from IMREAD_COLOR to IMREAD_UNCHANGED and adpat tests. The flag IMREAD_COLOR was always loading 3 channels, even for grayscale images while IMREAD_UNCHANGED adapts to the input file. See : https://docs.opencv.org/4.x/d8/d6a/group__imgcodecs__flags.html#ga61d9b0126a3e57d9277ac48327799c80

---

Implement the solution presented in issue #3 (closed).

• Change the constructor of TensorImpl to take the dimensions of the tensor as argument instead of the number of elements. For tensorImpl_opencv :
• Add mDims attribute in the class to use in Lazy_init (Instanciation of cv::Mat)
• Override the resize function from TensorImpl.hpp to update mDims

changed milestone to %aidge_backend_opencv - v0.1.0

requested review from @olivierbichler and @pineapple

assigned to @thibaultallenet

added 2 commits

• aa4a1656 - Change cv::Mat to cv::Mat*
• a6a1363d - Update tensorOpencv to avoid CopyCast with setDataType

Compare with previous version

changed title from Change tensorimpl opencv future_std::spancv::Mat to cv::Mat to Change tensorimpl opencv {++}future_std::span<cv::Mat>{++} to {++}cv::Mat{+*+}

changed the description

changed milestone to %aidge_backend_opencv - v0.0.2

added LanguageC++ label

added 1 commit

• a5f42002 - Change TensorImpl to store a cv::Mat & update tests accordingly

Compare with previous version

changed title from Change tensorimpl opencv future_std::span<cv::Mat> to cv::Mat{-*-} to Change tensorimpl opencv future_std::span<cv::Mat> to cv::Mat

changed the description

changed the description

added 1 commit

• c64f1a31 - Change opencv imread flag to IMREAD_UNCHANGED

Compare with previous version

marked the checklist item Change the opencv::imread flag from IMREAD_COLOR to IMREAD_UNCHANGED and adpat tests. The flag IMREAD_COLOR was always loading 3 channels, even for grayscale images while IMREAD_UNCHANGED adapts to the input file. See : https://docs.opencv.org/4.x/d8/d6a/group__imgcodecs__flags.html#ga61d9b0126a3e57d9277ac48327799c80 as completed

changed target branch from fix to dev

added 15 commits

• c64f1a31...2bc94a09 - 12 commits from branch dev
• ce2fe8c7 - merge rework_code into TensorImpl
• fca0a230 - Update TensorImpl constructor to take the tensor dimensions instead of the...
• d627b1d8 - Merge branch 'dev' into TensorImpl

Compare with previous version

added 1 commit

• b227ffee - Replace Stimuli -> Stimulus

Compare with previous version

changed the description

added 1 commit

• 1e3ce9e5 - Update colorFlag of cv::imread to UNCHANGED and fix tests on cv::Mat size.

Compare with previous version